Research On Depth Data Processing In Virtual View Drawing

With the advent of the information age,Virtual Reality(VR)and Augmented Reality(AR)appear in people's daily lives.The role played by in-depth data is also increasingly important.Before 2010,the acquisition of in-depth data required specialized equipment.The high prices and complex calculations of these devices made it impossible for deep data to be widely used in people's daily lives.In 2010,Microsoft introduced Kinect,the world's first consumer-grade depth camera that can obtain high-quality depth data and corresponding color images,and can be used in bone tracking,facial expression recognition,speech recognition and other fields.The wide range of applications Kinect has and the low price determine its good research and application prospects.The low cost of Kinect makes it widely used and affects the quality of the depth data it obtains.This limits the underlying applications for many applications.Therefore,how to improve the quality of deep data obtained by Kinect has become a hot topic in the current study.This paper analyzes the working principle of Kinect,and deals with the problem of edge cracks in the depth data acquired by Kinect,the edge misalignment of color images and depth images,and the absence of large depth data.The main tasks are as follows:1.Iterative adaptive median filtering algorithm.To solve the problem of cracks in the edge of objects in the depth data obtained by Kinect,an iterative adaptive median filter algorithm is proposed in this paper.Based on the correlation between the depth data,this algorithm adds a hollow point detection and iterative repair process in the median filtering process.It can well maintain the depth while successfully filtering out discrete voids and repairing cracks.Object edge information in the image.2.Hierarchical depth data correction algorithm.This article in-depth exploration of the principle of Kinect to obtain depth data,the analysis of the problem of the edge of the object in the depth image and the edge of the object in the color image is not aligned.The cause of the problem is found: the infrared light and natural light distortion rate is different and the depth of the camera and There is a distance between color cameras.For the problem of edge misalignment,this paper proposes a layer-based depth data correction algorithm.Its central idea is based on the correlation between the pixels in the depth image,using the histogram to layer the depth data,combined with the edge information of the color image to achieve Correction.The specific process is: through iterative adaptive median filtering and image morphology operations on the depth image to reduce the impact of tiny holes and cracks on subsequent operations;then the depth image is layered according to the histogram,with the help of color images.The edge of the object is weighted and averaged to correct the edge of the object of the depth image layer by layer,and then the layered depth image is merged according to a certain rule.Finally,the edge of the object in the depth image and the color image is consistent.Solve the following issues of deep data missing provides the basis.3.Adaptive depth data restoration algorithm.The lack of depth information is directly reflected in the deep image,which is the larger cavity,and how to solve the larger hole problem in the depth image is the problem of improving the quality of the depth image.In order to solve this problem,this dissertation presents a adaptive depth information restoration algorithm.To solve the smooth lack data area,this algorithm proposes to fill the missing depth data area with structural similarity as the weight coefficient.Against the nonsmooth lack of depth data area,this dissertation adds the idea of bilateral filter and the weighting coefficient of structure information to the fast marching algorithm.For the nonsmooth depth data missing area,this paper introduces the idea of bilateral filtering into the fast marching algorithm,and adds the structural information to the weight coefficient,to achieve the repair of the non-smooth depth data missing region.The adaptive depth data repair algorithm proposed in this paper maximizes the use of information in color images to achieve the filling of missing depth data and expands the range of depth data applications that Kinect acquires.In this dissertation,three algorithms are proposed for the three major problems in depth data acquired by Kinect.Through the analysis of the experimental results,the algorithm proposed in this paper has considerable improvements for the previous algorithm.